Fix from Dask bug in Datasets (#19409)

This commit is contained in:
Chen Shen 2021-10-15 15:04:52 -07:00 committed by GitHub
parent 07064cddf9
commit acfbf4c170
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -500,7 +500,7 @@ def from_dask(df: "dask.DataFrame") -> Dataset[ArrowRow]:
partitions = df.to_delayed() partitions = df.to_delayed()
persisted_partitions = dask.persist(*partitions, scheduler=ray_dask_get) persisted_partitions = dask.persist(*partitions, scheduler=ray_dask_get)
return from_pandas( return from_pandas_refs(
[next(iter(part.dask.values())) for part in persisted_partitions]) [next(iter(part.dask.values())) for part in persisted_partitions])